Why Shake Flask Results Don’t Always Translate
Shake flask results fail to predict bioreactor performance primarily because the two systems differ in oxygen supply, pH control, CO2 accumulation, and mixing intensity. The shake flask to bioreactor transition is the most common failure point in early bioprocess development, with up to 40% of first bioreactor runs producing significantly different growth or productivity compared to flask data.
Shake flasks rely on passive gas exchange through a cotton plug or membrane cap. There is no active dissolved oxygen (DO) control, no pH titration, and no ability to decouple mixing from oxygen transfer. In a stirred-tank bioreactor, each of these parameters is independently controlled. The result: a process optimised in flasks may experience oxygen limitation, pH shock, or shear damage upon transfer.
Understanding exactly which parameters diverge — and by how much — is the key to a successful shake flask to bioreactor transition. The sections below quantify each difference and show how to bridge the gap systematically.
Critical Parameters That Change During Transfer
Six parameters account for >90% of performance discrepancies between shake flasks and bioreactors: oxygen transfer (kLa), pH control, dissolved CO2, mixing time, shear stress, and temperature homogeneity.
| Parameter | Shake Flask (250 mL) | Bioreactor (2 L STR) | Impact on Culture |
|---|---|---|---|
| kLa (h-1) | 20–80 | 50–300 | O2 limitation if mismatched |
| OTRmax (mmol/L/h) | 5–20 | 15–80 | Growth rate ceiling |
| pH drift (units/batch) | 0.5–2.0 | <0.1 (controlled) | Metabolic shifts, product quality |
| dCO2 (% saturation) | 5–15 | <3 (stripped) | Growth inhibition at >10% |
| Mixing time (s) | 1–5 | 2–15 | Nutrient/pH gradients |
| Max shear (s-1) | 50–200 | 500–5,000 (tip) | Cell damage, morphology |
| Evaporation (%/day) | 1–5 | <0.5 (humidified gas) | Osmolality creep |
Matching Oxygen Transfer (kLa) Across Systems
Matching kLa between the shake flask and bioreactor is the single most reliable criterion for a successful process transfer. kLa represents the volumetric oxygen transfer capacity independent of driving force, making it directly comparable across vessel geometries.
Measuring kLa in Shake Flasks
Three methods are practical for measuring shake flask kLa:
- Sulphite oxidation method — dissolve 0.5 M Na2SO3 with 10-3 M CoSO4 catalyst; titrate residual sulphite at intervals. Chemical kLa overestimates biological kLa by 10–30% due to the enhanced reaction.
- RAMOS (Respiration Activity Monitoring System) — sealed flask with pressure sensor tracks headspace O2 depletion in real time. Gold standard for OTR measurement.
- Optical DO sensor spot — non-invasive patch (e.g., PreSens or PyroScience) on flask base. Use dynamic gassing-out: saturate with N2, resume shaking, fit DO recovery to kLa = -ln[(C* - CL)/(C* - C0)] / t.
Key Correlations for Shake Flask kLa
The maximum OTR in a standard Erlenmeyer flask follows the empirical correlation published by Maier and Büchs (2001):
Worked Example — Estimating Shake Flask kLa
Given: 250 mL unbaffled Erlenmeyer, 50 mL fill volume (VL/VF = 0.2), 220 rpm, 25 mm orbital diameter, 37 °C, water-like medium.
Estimate:
OTRmax ≈ 0.5 × d00.44 × n1.12 × VL-0.84 × d0.38
where d0 = 0.025 m, n = 3.67 s-1, VL = 50 × 10-6 m3, d = 0.085 m
Solving: OTRmax ≈ 12 mmol/L/h
At C* = 0.21 mmol/L (37 °C, air): kLa = OTRmax / C* ≈ 57 h-1
Bioreactor target: Set agitation and aeration in a 2 L STR to achieve kLa ≈ 55–60 h-1. For a dual Rushton turbine system at 0.5 VVM, this corresponds to approximately 300–400 rpm (reactor-specific — verify with the dynamic method).
Setting Bioreactor Conditions to Match
Once you know the shake flask kLa, use the Van’t Riet correlation for your stirred-tank bioreactor:
kLa = C × (P/V)a × vsb
Where C, a, b are system-specific constants (C ≈ 0.002–0.004, a ≈ 0.4–0.7, b ≈ 0.2–0.5 for ionic solutions in coalescing media). Adjust impeller speed (which sets P/V) and gas flow rate (which sets vs) iteratively to hit the target kLa.
OTR & kLa Estimator
Estimate kLa for both shake flasks and stirred-tank bioreactors using Büchs and Van’t Riet correlations. Compare across vessel sizes.
pH and CO2 Management Differences
pH divergence is the most underestimated source of failed transfers. In a shake flask, the only pH buffering comes from the medium itself (typically 20–50 mM phosphate or HEPES). As the culture grows, organic acid production (acetate, lactate) or ammonia release drives pH down or up by 0.5–2 units. The organism adapts, and you unknowingly optimise for uncontrolled pH conditions.
When you transfer to a bioreactor with tight pH control (±0.05), the culture faces a fundamentally different metabolic environment. An E. coli strain that reached OD 8 in a flask at final pH 5.5 may behave very differently in a bioreactor clamped at pH 7.0 — potentially growing faster but also producing more acetate due to higher glucose uptake rate.
Dissolved CO2 Accumulation
Shake flasks with foam plugs or silicone caps allow limited gas exchange. At high cell densities (OD > 5), headspace CO2 can reach 5–15%. Dissolved CO2 (dCO2) at these levels inhibits growth in mammalian cells (IC50 ≈ 120–150 mmHg) and alters the carbonate buffering equilibrium, causing pH depression even in buffered media.
In a sparged bioreactor, continuous air flow strips CO2 efficiently, maintaining dCO2 below 2–3%. If your flask-optimised culture relied on elevated CO2 as a (hidden) metabolic signal, it may behave differently once CO2 is stripped away.
Bridging Strategy
- Run a parallel flask with periodic offline pH measurement every 2–4 hours to map the actual pH trajectory.
- In the bioreactor, initially run with pH control OFF to observe natural pH drift. If it matches the flask profile, the organism’s metabolic state is consistent.
- Gradually introduce pH control — widen the dead-band first (±0.3), then tighten to ±0.1 over successive runs.
- For mammalian cells sensitive to CO2, consider adding 5% CO2 overlay in early bioreactor runs to match flask headspace conditions.
Developing a Transfer Protocol (Step-by-Step)
A structured transfer protocol reduces failed bioreactor runs from ~40% (ad hoc transfer) to <10% (systematic approach). The protocol below has four phases: characterise, match, bridge, and verify.
Phase 1: Characterise the Shake Flask Process
- Standardise flask conditions — fix flask volume (250 mL), fill ratio (20%), shaking speed (200–250 rpm), orbital diameter (25 mm), cap type (cotton plug or vented cap), and incubator CO2 (if applicable). Run all screening experiments under these exact conditions.
- Measure kLa — use one of the three methods above. Record at your standard shaking conditions. Expected range: 40–70 h-1 for 50 mL in 250 mL flask at 220 rpm.
- Profile pH — sacrifice parallel flasks at 2, 4, 6, 8, 12, 24 h (adjust to your culture duration). Measure offline pH. Plot the trajectory.
- Record growth kinetics — OD600/VCD at each timepoint. Calculate specific growth rate (μ) from exponential phase. Record final cell density and product titer.
Phase 2: Match Key Parameters in the Bioreactor
- Set kLa to match flask — use the dynamic gassing-out method to measure bioreactor kLa at several agitation/aeration combinations. Select the combination that gives kLa within ±15% of the flask value.
- Run without pH control — for the first bioreactor run, disable pH control. Let pH drift freely to observe whether the bioreactor pH trajectory matches the flask. This confirms that the metabolic state is consistent.
- Use identical medium and inoculum — same medium lot, same cell bank passage number, same inoculum density and volume ratio.
Phase 3: Bridge to Controlled Conditions
- Enable pH control with a wide dead-band (±0.3 units), then tighten to ±0.1 over 2–3 runs.
- Introduce a DO cascade (agitation → air flow → O2 enrichment) with a 30% DO setpoint.
- Test antifoam at 3 concentrations (0.005, 0.01, 0.02% v/v) and re-measure kLa after addition.
- Compare μ, final OD/VCD, and product titer to the Phase 2 run. Investigate any deviation >20%.
Phase 4: Verify Reproducibility
Run at least 3 replicates under the final bioreactor conditions. Accept the transfer if:
- Specific growth rate (μ) is within ±20% of the characterised flask value
- Final product titer is within ±25%
- Product quality attributes (e.g., glycosylation, charge variants) are comparable
- Coefficient of variation (CV) across replicates is <15%
Scale-Up Calculator
Compare 5 scale-up criteria side-by-side (P/V, tip speed, kLa, Re, mixing time) when sizing your first bioreactor runs.
Common Pitfalls and How to Avoid Them
Even with a structured protocol, specific technical mistakes derail the shake flask to bioreactor transition. The following pitfalls account for the majority of failed transfers.
1. Overfilling Shake Flasks
Filling a 250 mL flask with 100 mL (40% fill ratio) instead of 50 mL (20%) reduces kLa by 60–70%. The culture becomes oxygen-limited, grows at a sub-maximal rate, and you set expectations for a slow process that won’t reproduce in a well-aerated bioreactor. Your “optimised” process was actually optimised for oxygen limitation.
2. Antifoam Shock
Bioreactors generate foam from sparging; shake flasks do not. Adding antifoam (silicone- or polypropylene glycol-based) above 0.01% v/v reduces kLa by 15–50% by increasing bubble coalescence and reducing interfacial area. If you set your bioreactor aeration to match flask kLa before adding antifoam, you end up oxygen-limited once antifoam is introduced.
Solution: Always measure kLa with antifoam present at the concentration you intend to use in production. Alternatively, use mechanical foam destruction (impeller above liquid level or mechanical breaker) to minimise chemical antifoam.
3. Ignoring the pH Trajectory
A flask culture of E. coli on glucose that drifts from pH 7.0 to 5.5 over 12 hours is experiencing progressive acid stress. The culture adapts — it upregulates acid-tolerance genes (RpoS regulon), reduces growth rate, and may shift metabolic flux. When you transfer to a bioreactor clamped at pH 7.0, you remove this selection pressure. Growth rate may increase but acetate production often rises because the specific glucose uptake rate is no longer limited by acid stress.
4. Different inoculum preparation
Inoculating a bioreactor from a single flask that was prepared differently from your screening flasks (different passage number, different pre-culture duration, different growth phase at harvest) introduces variability that has nothing to do with the vessel. Always use identical pre-culture protocols.
5. Ignoring evaporation effects
Shake flasks lose 1–5% of their volume per day to evaporation (depending on cap type and incubator humidity). Over a 48-hour culture, this concentrates all solutes by 2–10%, increasing osmolality. Bioreactors with humidified inlet gas and water-jacketed exhaust condensers lose <0.5% per day. Your flask-optimised medium osmolality may be too high for bioreactor conditions where evaporative concentration does not occur.
| Pitfall | Symptom in Bioreactor | Root Cause | Solution |
|---|---|---|---|
| Overfilled flasks | Faster growth than expected | Flask was O2-limited; bioreactor is not | Standardise fill ratio to 10–20% |
| Antifoam shock | DO drops, culture limited | kLa reduced 15–50% by antifoam | Measure kLa with antifoam present |
| pH mismatch | Different metabolic profile | Flask drifted; bioreactor is controlled | Map flask pH; bridge gradually |
| CO2 stripping | Higher growth rate, different product quality | Flask CO2 was inhibitory; bioreactor strips it | Add CO2 overlay in early runs |
| Shear damage | Lower viability, morphology change | Impeller tip speed too high | Start at lower rpm; use pitched blade |
| Evaporation | Lower osmolality in bioreactor | Flask concentrated solutes; bioreactor did not | Adjust media osmolality for bioreactor |
Gas Mixing Calculator
Calculate gas blending ratios for bioreactor aeration — air, O2, N2, and CO2 mixtures with flow rate conversions.
Frequently Asked Questions
What is the typical kLa range in a shake flask vs a bioreactor?
Standard 250 mL Erlenmeyer flasks at 200–250 rpm with 50 mL fill volume deliver kLa values of 20–80 h-1. A 2 L bench-top stirred-tank bioreactor at 1–3 W/L and 0.5–1 VVM typically achieves 50–300 h-1. The gap narrows at low oxygen demand (mammalian cell culture, kLa 5–15 h-1) but becomes a major bottleneck for aerobic microbial fermentations requiring kLa above 100 h-1.
Why does my E. coli grow slower in the bioreactor than in shake flasks?
Common causes include: higher shear stress from Rushton turbines damaging cells at the impeller tip, excessive antifoam addition suppressing kLa by 10–50%, uncontrolled pH drop from organic acid accumulation (shake flasks buffer passively with headspace CO2 exchange), and lower effective DO if the bioreactor aeration rate was not matched to the culture’s oxygen demand. Always profile DO and pH in both systems before concluding the bioreactor is at fault.
Should I match kLa or OTR when transferring from shake flask to bioreactor?
Match kLa as the primary criterion. OTR depends on the driving force (C* − CL), which changes with DO setpoint and back-pressure. Matching kLa ensures equivalent oxygen transfer capacity regardless of the operating DO level. Measure shake flask kLa using the sulphite method or a RAMOS device, then set bioreactor agitation and aeration to achieve the same kLa using Van’t Riet or your reactor’s empirical correlation.
How do I measure kLa in a shake flask?
Three practical methods: (1) The sulphite oxidation method — add sodium sulphite with cobalt catalyst and measure depletion rate; gives a chemical kLa that slightly overestimates biological kLa. (2) A Respiration Activity Monitoring System (RAMOS) that measures oxygen transfer rate from headspace O2 depletion in real time. (3) A non-invasive optical DO sensor spot (e.g., PreSens) adhered to the flask base — use the dynamic gassing-out method (sparge N2, then resume shaking and fit the DO recovery curve to get kLa).
What fill volume ratio should I use in shake flasks for scale-up data?
Use 10–20% of nominal flask volume (e.g., 25–50 mL in a 250 mL flask) for aerobic cultures. Higher fill volumes (>25%) drastically reduce kLa because the liquid cannot form the thin surface film needed for efficient gas–liquid transfer. For reliable process transfer data, standardise your fill ratio and shaking speed, and always report both alongside growth data.
Related Tools
- OTR & kLa Estimator — Estimate kLa for both shake flasks (Büchs correlation) and stirred tanks (Van’t Riet). Compare vessel sizes side-by-side.
- Scale-Up Calculator — Calculate target RPM for 5 scale-up criteria (P/V, tip speed, kLa, Re, mixing time) across any two vessel sizes.
- Gas Mixing Calculator — Blend air, O2, N2, and CO2 to hit target inlet compositions for bioreactor aeration.
References
- Büchs J. (2001). Introduction to advantages and problems of shaken cultures. Biochemical Engineering Journal, 7(2), 91–98. doi:10.1016/S1369-703X(00)00106-6
- Klöckner W, Büchs J. (2012). Advances in shaking technologies. Trends in Biotechnology, 30(6), 307–314. doi:10.1016/j.tibtech.2012.03.001
- Schulte A, Jordan A, Klöckner W, Schumacher M, Corves B, Büchs J. (2025). Effect of high-speed shaking on oxygen transfer in shake flasks. Biotechnology Journal, 20(4), e70013. doi:10.1002/biot.70013
- Brauneck G, Engel D, Grebe LA, Hoffmann M, Lichtenberg PG, Neuß A, Mann M, Magnus JB. (2025). Pitfalls in early bioprocess development using shake flask cultivations. Engineering in Life Sciences, 25(1), e70001. doi:10.1002/elsc.70001